大工至善|大学至真分享 http://blog.sciencenet.cn/u/lcj2212916

博文

[转载]【雷达与对抗】【2019.09】利用Landsat时间序列绘制安纳托利亚东南部冬春作物图

已有 142 次阅读 2020-3-25 16:21 |系统分类:科研笔记|关键词:学者|文章来源:转载

本文为柏林洪堡大学地理研究所的学士论文,共45页。

 

为了更好地了解农业系统对土地和水资源的需求,对农田土地利用强度进行监测至关重要。通过对时间序列的分析,遥感可以提高对农田系统时空变化的认识。安纳托利亚东南部项目(GAP)位于半干旱的幼发拉底河底格里斯流域,是土耳其规模最大、成本最高的区域开发项目,旨在储存和分配幼发拉底河和底格里斯河的水,用于灌溉耕地。为了估计冬季和春季作物在GAP中的空间分布,本研究在2015年使用了140个经过大气校正和地形归一化的Landsat收集的12WRS-2足迹的1级图像,计算了季度合成和光谱-时间度量。为了绘制冬春作物图,收集了训练数据,并将光谱时间变异性指标应用于201541日至626日的三个月合成的非参数分类器,计算面积调整精度以评估地图精度、面积和95%置信区间。结果表明,冬、春两季作物分类的总准确率分别为83.71%(±2.31%)、89.23%(±4.13%)和97.58%(±2.35%)。经耕地面积调整计算,2015年冬春季作物总面积为1737355公顷(±88639公顷),相当于研究区域的22.2%(±1.13%)。但由于与草地面积之间的混淆,相对较高的遗漏误差(10.77%)导致冬春作物等级估计偏低。省一级的统计数据显示,Diyarbakır415739公顷)和Şanlıurfa431525公顷)的冬、春作物得到绝对最大的扩展,而KilisAdıyamanDiyarbakırBatmanSiirt省的冬、春作物在农田上的比例最高(大于80%)。假设有足够的数据可用,这种方法可以用来绘制冬春季作物的年度变化图。

 

It is crucial to monitor land use intensityof cropland in order to better understand the land and water resource demandsof agricultural systems. Remote sensing can improve the understanding ofspatial and temporal variabilities of cropland systems through the analyses oftime series. The Southeastern Anatolia Project (GAP), located in the semi-aridEuphrates-Tigris Basin, is Turkey’s largest and most expensive regionaldevelopment program, aimed to store and distribute water from the Euphrates andTigris rivers for the irrigation of arable land. To estimate how winter andspring crops are spatially distributed in the GAP, this study used 140atmospherically corrected and topographically normalized Landsat Collection 1Tier 1 images across 12 WRS-2 footprints in 2015, to compute a quarterlycomposite and spectral-temporal metrics. To map winter and spring crops,training data was collected and with spectral-temporal variability metricsapplied to a non-parametric classifier for the three-month composite from 1stApril until 26th June 2015. Area-adjusted accuracies were calculated to assessmap accuracy, area and 95% confidence intervals. The results showed asatisfactory separation of land cover, reaching an overall accuracy of 83.71 (±2.31%), producer’s accuracy of 89.23% (± 4.13%) and user’s accuracy of 97.58%(± 2.35%) for the winter and spring crops class. Cropland areaadjustedcalculations revealed a total expanse of winter and spring crops of 1,737,355hectares (± 88,639 ha) of land in 2015, which is equal to 22.2% (± 1.13%) ofthe study area. However, the relatively high error of omission (10.77%)proposes an underestimation of the winter and spring crops class due to theconfusion with the grassland class. Province-level statistics detected theabsolute largest expansions of winter and spring crops in Diyarbakır (415,739ha) and Şanlıurfa (431,525 ha), while the highest proportion (>80%) ofwinter and spring crops on agricultural land is found in the provinces ofKilis, Adıyaman, Diyarbakır, Batman and Siirt. The approach could betransferred to create annual change maps for winter and spring crops, assumingsufficient data availability.

 

1. 引言

2. 研究领域

3. 数据与方法

4. 结果

5. 讨论

6. 结论

7. 致谢

附录


更多精彩文章请关注公众号:qrcode_for_gh_60b944f6c215_258.jpg



http://blog.sciencenet.cn/blog-69686-1225249.html

上一篇:[转载]【统计学】【1999】基于归纳推理技术的时间序列预测
下一篇:[转载]【电力电子】【2013】【含源码】考虑馈电线阻抗的三相交直流变换器建模与仿真

0

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...
扫一扫,分享此博文

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2020-6-6 07:22

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部